This post originated from an RSS feed registered with Scala Buzz by Caoyuan Deng. Original Post: Progress of Scala for NetBeans - with a New Written Lexer and Parser Feed Title: Blogs about Scala from Caoyuan Feed URL: http://blogtrader.org/page/dcaoyuan/feed/entries/atom?tags=scala Feed Description: Blogs about Scala from Caoyuan Deng Latest Scala Buzz Posts Latest Scala Buzz Posts by Caoyuan Deng Latest Posts From Blogs about Scala from Caoyuan

According to this post:

I talked to Jetbrains about this, and they told me that they stopped working on the Scala plugin for the time being, because - demand for Groovy/Ruby was higher - the language was moving too fast - Scala is a terribly difficult language for compiler/tool writers, and the only good way to analyze Scala programs might be through the official compiler, which didn't yet support this

It's true that "Scala is a terribly difficult language for compiler/tool writers", but I'm trying to bypass "the only good way to analyze Scala programs might be through the official compiler"

Before rewriting Scala for NetBeans, I considered some parser choices, one was Scala's native compiler, which is good for compiling/building Scala project, but not suitable for Editor. And JavaCC, ANTLR, which may be good enough, but it's not natural to express Scala's grammar.

Then I found Rats! which is used by Fortress, a very very clean, powerful parser generator. After couple of days working, I got an incrementally lexer for Scala, and a parser for Scala that with Scala's grammar being naturally expressed (the grammar definition is ParserScala.rats). The benefit of a complete controllable parser is that I can now do some type inference and wholly semantic analysis freely and immediately.

Another progress is that I've decoupled the Scala project's dependency on Java's classpath in NetBeans, instead, GSF's classpath is used in Scala project module now. That means, I can begin the indexer for Scala's standard library and project source files.

The next steps will be type inference; smart completion with type inferred information; indexer for later refectory and usages searching; parsing error recover etc.